Skip to main content

Full text of "The Cause of Luminosity in the Flames of Hydrocarbon Gases"

See other formats

450 Prof. V. B. Lewes. The Cause of Luminosity [Mar. 21^ 

IS'ovember 2, the flow was never less than 3,000 millions of gallons 
per day, and on November 2 it reached 4,240 millions. Again, on 
the 17th the flow was 3,305 millions, and on the 18th, 4,165 millions. 
It then gradually decreased to 1,845 millions on the day when the 
sample was drawn. Thus the Thames basin had been twice very 
thoroughly washed out immediately before the time when the 
November sample was taken. There had previously been no such 
fl.oods after the 5th of January in that year. This condition of things 
affords a fairly satisfactory explanation of the anomalous result 
yielded by this sample. 

III. *' The Cause of Luminosity in the Flames of Hydrocarbon 
Gases." By ViVlAi^^ B. Lewes, Professor of Chemistry at 
the Royal Naval College, Greenwich. Communicated by 
Pi'ofessor Thorpe, F.R.S. Received February 14, 1895. 

In a paper read before the Chemical Society in 1898, I showed 
that in the inner non-luminous zone of aflame of ordinary illuminating 
gas, the hydrocarbons originally present in the gas, and consistiug of 
ethylene, butylene, benzene, methane, and ethane, became converted by 
the baking action of the walls of flame between which they had to pas& 
into acetylene, and that at the moment when luminosity commenced,, 
over 80 per cent, of the total unsaturated hydrocarbons present 
consisted of this compound. 

The presence of acetylene at the point where luminosity commenced 
naturally suggested that it was in some way due to actions in which 
the acetylene played the principal part — either that it split up intO' 
carbon and hydrogen under the influence of heat, and so supplied the 
flame with the solid particles necessary, according to Sir Humphry 
Davy's theory of the cause of luminosity, or else by its polymerisation 
it formed the dense vapours required by Dr. E. Frankland's more 
recent hypothesis. 

In order to elucidate this point, I carried out the long series of 
experiments upon the action of heat upon flowing ethylene and other 
liydrocarbons, which formed the subject of communications to the 
Royal Society in 1893 and early this year, in which I showed that 
whilst flowing through a heated area (the temperature of which was 
between 800° and 1000° C), ethylene decomposed according to the 

ec^Hi = 2C2H2+2CH4, 

and that the acetylene then polymerised into a large number of more 
complex hydrocarbons, amongst which benzene and naphthalene were 
conspicuous, whilst at temperatures above 1200° C, no polymerisation 

1895.] in the Flames of Hydrocarhon Gases. 451 

took place, but the acetylene formed from the ethylene decomposed 
at once into carbon and hydrogen, whilst the methane, which up to 
this temperature had been but little affected, decomposed into 



and this fresh supply of acetylene at once broke up to carbon and 
hydrogen, so that at temperatures above 1200° 0. the complete 
action might be looked upon as being 

CK, = a + 2H,. 

These results have an important bearing upon the cause of the 
luminosity in the flame, as it is manifest that if the temperature 
of the luminous zone is above 1200° C, the light emitted must be 
due to incandescent particles of carbon, and not to incandescent 
hydrocarbon vapours. 

On determining the temperature of an ethylene flame whilst burn- 
ing from a small fish-tail burner by means of the Le Chatelier thermo- 
couple, used in the way described in my paper* on the luminosity of 
coal-gas flames, I found that the temperatures were as follows : — 

Height above 
Portion of flame. burner. Temperature. 

Non-luminous zone |- inch. 952° C. 

Commencement of luminosity ... 1^ ,5 1340 

Top of luminous zone 2 „ 1865 

Sides of „ ,, c 1875 

showing that luminosity commenced at 1340° C, and continued even 
at 1875° C, temperatures at which the incandescent vapour theory 
becomes untenable. 

It might be urged that the heavy hydrocarbons already produced 
at a lower temperature in the non-luminous zone are not so easily 
decomposed by heat as acetylene, and that these may be causing the 
luminosity, even though carbon particles be present from the decom- 
posed acetylene ; but this would hardly be possible, as so little besides 
acetylene is to be found at the top of the non-luminous zone of an 
ethylene flame, and it can be experimentally shown that even when 
benzene vapour is formed and is largely diluted it begins to break up 
and deposit carbon at 1200° C. 

The supporters of the " solid particle " theory of luminosity agree 
in concluding that the liberated carbon, existing as it does in a con- 
dition of molecular division, is heated to incandescence partly by its 
own combustion, and partly by the combustion of the hydrogen and 
carbon monoxide going on around the finely-divided carbon particles. 

* ' Chem. Soc. Journal/ 1893. 

2 K 2 

452 Prof. V. B. Lewes. The Cause of Luminosity [Mar 21, 

As lias been pointed out by many observers, it is clear tbat tbe 
carbon particles themselves nndergo combustion, otherwise they v^ould 
escape nnburnt from the flame, whilst it is manifest that the com- 
bustion of hydrogen and carbon monoxide, which plays so important 
a part in the flame, must add its iota to the temperature attained by 

Both these sources of temperature, however, v^ould be manifest in 
the flame itself, and v^ith flames of given size burning from the same 
description of jet we ought to find that their luminosity is governed 

A. The temperature of the flame. 

B. The number of carbon particles in a given area. 

Moreover, we should expect that the higher the temperature of the 
flame, the v^hiter would be the light emitted, so that a comparatively 
low temperature flame, even when rich in carbon particles, w^ould 
be yellow and lurid as compared vv^ith a flame containing the same or 
a smaller number of particles, but which had a higher temperature. 

It has been pointed out by Professor A. Smithells* that it is 
erroneous to consider the temperature of a flame as being the tem- 
perature recorded by thermometric instruments inserted into the 
flame, as by such devices you only obtain the mean temperature of a 
considerable area of the flame uncorrected for loss from conduction. 

It is also perfectly vrell known that in a flame a thick platinum 
wire may only be heated to redness, whilst a thin wire may even be 
fused, and this suggests that flame temperatures taken by the Le 
Oh atelier thermo-couple of platinum and platinum-rhodium wires 
may be totally incorrect. In using this beautiful and convenient 
device, I have found that the length of the wires twisted together 
made piactically no difference in the recorded temperature, and that 
one twist was as good as six. 

In all my flame experiments I have made the twist as short as 
possible, and by always using wires of the same thickness have 
obtained results which are at any rate comparable if not correct, and 
in order to find what difference the thickness of the wires would make, 
I got Messrs. Johnson and Matthey to draw for me wires of 0*018, 
O'Oll, and 0"003 of an inch diameter, and having calibrated the 
galvanometer scale for temperature with thermo-couples of the same 
length of twist made from each of them, obtained the following 
results with the same portion of a Bunsen flame. 

Wire used. Temperature shown. 

0-018 1617° C. 

0-011 1728 

0-003 1865 

^ * Phil. Mag.,' 1894, p. 249. 

1895.] in the Flames of Hydrocarbon Gases, 453 

These results show that the diameter of the wire seriously affects 
the temperature recorded under these conditions by the thermo- 
couple, the same degree of heat being recorded by the fine wire as 
being 248° hotter than is shown by the thickest wire employed, this 
discrepancy being probably chiefly due to loss by conduction. 

In taking the temperature of heated gas flowing through a tube 
this source of error is but small, as some considerable length of wire 
being heated on each side of the twist, conduction has but little effect 
on the thermo-couple itself, but in determining the temperature of 
flames it is manifest that the finest usable wire must be employed in 
order to reduce the error from conduction. Test experiments also 
showed that no part of the thermo-couple must project beyond the 
flame, as if it did a considerable diminution in the recorded temperature 
took place. 

For these reasons it was manifestly best to use the finest wire 
which could be employed without the risk of fusing at the tempera- 
tures existing in the flames to be tested ; and all temperatures 
recorded in this paper were made with wire 0*011 in. in diameter, 
thfe twist being as short as possible, so that it is probable that, 
although the temperatures may be from 100° to 200° too low, yet the 
results are strictly comparable. 

Experiments which I have lately made with pure acetylene, prepared 
by the action of water upon calcic carbide, show it to be the most 
powerful illuminant to be found amongst the gaseous hydrocarbons, 
as when burnt in a small flab flame burner under the most suitable 
pressure, and its illuminating power calculated to a flow of 5 cubic ft. 
an hour, its value is equal to about 240 candles. 

The colour of the flame is pure white, and an ethylene flame beside 
it looks yellow and dull — the purity of the light at once suggesting 
a very high condition of incandescence in the particles of carbon 
present in the flame. 

On now taking the temperature of the various portions of the 
flame, and comparing these with the temperatures obtained in the 
same way with the ethylene flame and a coal-gas flame of the same 
size, the following results are obtained :— - 

Portion of flame. Acetylene. Ethylene. Coal gas. 

I^on-luminous zone 459" C. 952° 0. 1023° 0. 

Commencement of luminosity. . 1411 1340 1658 

Near top of luminous zone .... 1517 1865 2116 

whilst the illuminating values of the gases calculated to a flow of 
5 cubic ft. an hour in the burners best suited for their consumption, are 

Acetylene 240*0 

Ethylene , 68*5 

Coal gas 16*8 

454 Prof. V. B. Lewes. The Cause of Luminosity [Mar. 21. 

whilst if all were compared when burning from flat-flame burners 

of the same size as those in which the temperatures were determined, 

the results when calculated to a consumption of 5 cubic ft. an hour 

would be 

Acetylene 211'0 

Ethylene 31-5 

Coal gas nil 

Here then we have the anomaly of three gases, which not only do 
not conform to the preconceived expectation, but which have their 
ratio of temperature and illuminating value directly opposed to each 

In the case of the acetylene and ethylene, moreover, the molecules 
contain the same number of atoms of carbon, and yet we obtain so 
enormous a discrepancy in their illuminating value. 

The fact that there is no apparent relation existing between the 
temperature of the flame, or the probable number of carbon particles 
contained in it and its illuminating value, at once suggests that the 
luminosity must be in great part governed by some thermo-chemical 
changes taking place in the flame itself, and which do not of necessity 
affect the average temperature of the flame to any great degree. 

The researches of Hittorf* and Siemens show that air, steam, and 
the oxides of carbon, even when heated to temperatures above those 
existing in luminous hydrocarbon flames, are perfectly non-luminous, 
and the fact that the Bunsen flame, when supplied with safficient air, 
has a temperature exceeding 1800° 0. in its hottest part, and yet 
emits no light, shows us that it is exceedingly unlikely that any 
interactions leading to luminosity take place amongst these ordinary 
flame gases. 

The fact that most of the unsaturated hydrocarbons in the flame 
are converted into acetylene before luminosity commences, naturally 
draws one's attention to this body, and the fact that it is highly 
endothermic, at once suggests the idea that it may be the liberation 
of heat during its decomposition that endows the carbon particles 
produced from it with an incandescence far higher than any which 
could be expected from the temperature of the flame. 

Berthelot has calculated that the temperature developed by the 
detonation of acetylene at constant volume is no less than 6220° C, 
and if this be imparted at the moment of its liberation to the products 
of its decomposition, the incandescence of the carbon particles is at once 

If luminosity be even partly due to this cause, the detonation of 
pure acetylene first recorded by Berthelot should develop light. In 
order to see if this were so, a thin glass tube, closed by a cork, had 

* ' Wied. Ann.,' vol. 7, pp. 587, 591. 

1895.] in the Flames of Hydrocarhon Gases. 455 

a detonator containing one-tentli of a gram of mercuric fulminate 
suspended in it by two copper wires, wMcli were connected by a thin 
platinum wire in contact witb tbe fulminate, and on firing the 
detonator by the electric current the flash, of the fulminate was 
found to emit but a feeble light. 

Tbe same charge was fixed in a similar tube filled with, pure 
acetylene collected over mercury, the result being a flash of intense 
wbite light and th.e shattering of the tube, th.e pieces of which, were 
thickly coated with, the carbon produced by the decomposition of the 

Moreover, the small piece of white tissue paper used to contain the 
fulminate was only scorched at the points where the explosion of the 
fulminate had burst througb it, showing that in the instantaneous 
decomposition which had taken place, th.e intense beat which had 
been developed either was confined to the products of decomposition, 
or else bad not had time to scorch the paper. 

The experiment at first sight seemed conclusive evidence that it 
was the endothermic nature of the acetylene which, during its decom- 
position in the flame, endowed the particles of carbon with the 
necessary incandescence, but the objection presented itself that, when 
exploding mixtures of oxygen and hydrogen in the eudiometer, a 
distinctly luminous flash is produced, and, although the light so 
obtained is feeble as compared with the intensity of the white light 
produced by the detonation of the acetylene, still further proof is 
necessary before this action can be accepted as the prime factor in 
producing luminosity. 

It is also manifest that it would not do to assume that the rapidity 
of the decomposition of the acetylene in a flame was nearly so great 
SiS when the undiluted gas was detonated, and the question arose as 
to whether it would be possible to obtain evidence as to acetylene, 
when exposed to heat alone, liberating carbon in a luminous con- 

Although the instantaneous liberation of heat on the decomposition 
of the gas by detonation appears to confine the temperature to the 
products of its decomposition, it was to be expected that, on being 
decomposed by heat, and probably, therefore, at a slower rate, the 
increase in temperature might be detected. 

To try this, pure acetylene was passed through a platinum tube 
-2 mm. in diameter and 40 cm. long, in which the Le Chatelier 
thermo-couple was arranged as follows :— The two wires were twisted 
together for a length of 3 mm., and the wires on either side of the 
twist are then passed through thin glass tubes, which are fused on to 
them ; having been in this way coated with glass so that only the 
twist is exposed, they are passed through the platinum tube, the 
glass insulating the wire from the metal of the tube, and also keeping 

456 Prof. V. B. Lewes. The Cause of Luminosity [Mar. 21 5. 

the thermo-jnnction in sucli a position that it registers the tempera- 
tares of the gas in the tube, not that of the wall of the tube. To 
each end of the platinum tube glass "[""Pi^^^^ ^^^ fitted, down the 
steins of which the wires pass to mercury seals ; from tbe metal seals 
conducting wires lead to the resistance coils, the key, and a reflecting^ 

A sfceady flow of acetylene was allowed to pass through the tube,, 
and was led into water at the other end. The tube was slowly and 
carefully heated for about 4 in. of its length, and, as the tempera- 
ture reached 700'' C, white yapours began to flow from the tube, 
and these, as the temperature rose, increased in quantity. The 
source of heat had been so regulated that the temperature had risen 
about 10° per minute, but, almost immediately 800° 0. was passed, the 
galvanometer registered a sudden leap up in temperature to about 
1000° C, whilst finely- divided carbon poured from the tube. This 
seemed to indicate that 800° was about the temperature at which the 
pure acetylene broke up into its constituents, and an experiment was 
now made to see if this developed incandescence in the liberated 

A small glass combustion tube was well supported, and heated ta 
the highest temperature attainable with one of Fletcher's big blow- 
pipes, whilst pure acetylene was slowly flowing through it, the 
heating not being commenced until the tube was filled with the pure 
gas, all air being thoroughly rinsed out. As the temperature reached 
the softening point of the glass, the acetylene apparently burst into a 
lurid flame at the point where it entered the zone of heat, and clouds 
of carbon swept forwards through the tube ; but, although the carbon 
particles had to traverse an inch or more of tube more highly heated 
than the point of entering the hot zone, it was only at this latter 
point that the luminosity was developed, proving beyond doubt that 
it was the heat evolved by the decomposition, and not the external 
heating, which caused the carbon particles to emit light. 

If it is fche decomposition of the molecule of acetylene which 
develops the heat which is the cause of the incandescence of the 
carbon particles, then, if acetylene could be burnt without decomposi- 
tion, a non-luminous flame should be produced. It is conceivable 
that this might be done by so diluting the acetylene that it would 
require a much higher temperature to break it up. 

It was Heumann who showed* that hydrocarbon gases may burn 
with luminous flames, i.e., with separation of carbon in the flame, or 
with non-luminous flames, i.e.^ without any separation of carbon, and 
that the maintenance of a high temperature is an essential condition 
of luminosity : a flame, the temperature of which has been lowered 
by any means, being no longer able to bring about the required 

* ' Liebig's Annalen,' toI. 183, Part I, pp. 102—131. 

1895.] in the Flames of Hyclrocarhon Gases. 457 

separation of carbon. He also points out* that " combnstible matter, 
wlien diluted with indifferent gases, requires to "be maintained at a 
higher temperature, in order that it may burn with a luminous flame,; 
than when it is undiluted with such gases." 

Dr. Percy JFrankland, in his researches on the effect of diluents 
upon the illuminating yaiue of hydrocarbons, f showed that ethylene, 
which was capable of developing a light of 68*5 candles power when 
burnt by itself, became non-luminous when diluted with about :— 

Hydrogen 90 per cent. 

Carbon monoxide ....,*., 80 ,, 

Carbon dioxide .......... 60 ,, 

Nitrogen «».... 87 „ 

results which all show that excessive dilution by inert gases destroys 

In order to see if dilution had the same effect upon acetylene,, 
experiments were made by diluting it with pure hydrogen. The 
gases were mixed over water, the proportion of acetylene actually 
present in the gas being determined by analysis at the burner, and 
although the water in both holder and meter was, as far as possible^ 
saturated with the gas, yet, as the analyses show, the precaution was 
an important one. 

Composition of mixture, 



value of mixture 

per 6 c.c. when 

burnt in 00 Bray. 



Made in holder. 

' A . 

At bnrn 

/ - -^ — 



— \ 























Showing that dilution with between 80 and 90 per cent, of hydrogen 
rendered the acetylene non-luminous when the mixture was burnt 
from a burner suitable for the higher values of gas. 

In order to determine the point at which luminosity was destroyed 
when consuming the mixture in a burner suited to develop the light 
from a gas of low illuminating power, the experiment was repeated, 
using a 3-in. flame burning from the London argand, and also from a 
JN^o. 4 Bray union jet, the latter being employed as it is difficult to 
determine the temperature in the argand flame. 

* * Liehig's Annalen,' vol. 183, Part I, pp. 102 — 131. 
t ' Chem. Soc. Jour.,* vol. 45, p. 30 and p. 227. 

458 Prof. V. B. Lewes. The Cause of Luminosity [Mar. 21, 

Illuminating ralue 
Analysis of mixture. per 5 cub. ft. 

Hydrogen. Acetylene. Argand. IS'o. 4 Bray. 

92 8 JSTot measurable 

91 9 ISTot measurable 

5 14-5 4-1 17 

so that luminosity would be destroyed in tbe argand by dilution with 
about 90 per cent, hydrogen, and in the Ko. 4 Bray with about 88 
per cent. 

The next point to be determined was whether the destruction of 
luminosity in the diluted acetylene flame was in reality due to dilu- 
tion rendering it necessary to employ a higher temperature for the 
decomposition of the acetylene, or to other causes. 

In order to do this, a tube made of specially infusible glass 4 mm. 
in diameter was taken, and the Le Ohatelier thermo-couple was 
fitted into it in the same way as before, used with the platinum tube, 
and all air having been rinsed out by a current of the mixture to be 
experimented with, the gas was allowed to pass at a steady rate of 
flow through the tube, the point at which the thermo-couple was 
situated being steadily heated by the Fletcher blowpipe, whilst the 
temperature recorded on the scale was noted the moment that incan- 
descent liberation of carbon commenced. 

Percentage composition of gas. Temperature necessary 

^ A ^ ^Q cause deposition of 

Acetylene. Hydrogen. carbon with luminosity. 

100 780^0. 

90 10 896 

m 20 1000 

It was found impossible to obtain a glass tube which would stand 
temperatures higher than this ; but on plotting out the points so 
obtained, and which give a fairly straight line, it is seen that even if 
the increase in temperature only continues for increased dilution in 
the same ratio as shown in the experimental determinations, which 
is extremely unlikely, the reason of the destruction of luminosity in 
highly-diluted hydrocarbon gases is at once explained, as an increase 
of each 10 per cent, in the dilution would necessitate an increase of 
100° 0. in the temperature of the flame, and with 90 per cent, dilu- 
lution a temperature of over 1700° C. would be required to bring 
about decomposition. 

My reason for believing that it is highly improbable that when 
dilution is great it only requires the same increment in tepiperature 
to bring about decomposition as when the dilution is small, is that in 
all the work I have done on the effect of diluents upon luminosity, 
and also in Professor Percy Prankland's researches upon the same 


m the Flames of Hydrocarbon Gases, 


¥m. 1. 

subject, dilution witli hydrogen and carbon monoxide acts regularly, 
and decreases the value of the illuminant in a direct ratio down to 
about 60 per cent., whilst when the degree of dilution exceeds 60 per 
cent, a rapid falling away in the luminosity takes place, a fact which 
I think points clearly to a regular pro rata rise of temperature being 
needed for increase in dilution up to between 50 and 60 per cent., 
whilst higher degrees of dilution need a far greater rise of tempera- 
ture in order to bring about decomposition. 

Moreover it would be manifestly incorrect to look upon the per- 
centage of acetylene present in the gas issuing from the burner as 
being any guide to the degree of dilution existing at the point at 
which luminosity commences. As the two small streams of gas issuing 
from the holes in the union jet meet and splay themselves out into 
the flat flame, they draw in with them a considerable proportion of 
air, the quantity being governed by the pressure of the gas at the 

This can be clearly seen by the fact that a high value gas which 
burns from a union jet burner of a given size with a smoky flame, 
under a gas pressure of half an inch of water, will burn with a 
bright, smokeless, and rigid flame of greatly increased illuminating 
value when the pressure is raised to 2 in,, whilst an ordinary coal 
.gas of 16-caudle value must be burnt from a flat flame burner at a 
pressure of about 0*75 in. if the best results are to be obtained, the 
increase in air drawn in, if the pressure rises to a much higher degree, 
diminishing the illuminating value. 

Then, again, the area of non-luminous combustion in a mixture of 
gases like coal gas means that some at least of the hydrocarbons are 
consumed before the required temperature for their decomposition is 

460 Prof. V. B. Lewes, The Cause of Luminosity [Mar. 21^ 

reached, whilst the prodncts of combustion formed in the lower part 
of the flame are mixed with the flame gases, partly bj diffusion and 
partly by being drawn into it by the upward rush. 

When a simple hydrocarbon like ethylene or acetylene is burnt 
alone, the whole of the heat required to bring about the decomposi- 
tion has to be generated by the combustion, without decomposition, 
of a considerable proportion of the hydrocarbon, and this means con- 
siderable dilution at the spot where the luminosity commences, sa 
that at the top of the non-luminous zone of an acetylene flame there 
is only some 14 or 15 per cent, of acetylene present, diluted with 
nitrogen, hydrogen, water vapour, and the oxides of carbon, whilst, 
with a mixture of 10 per cent, acetylene and 90 per cent, of hydrogen, 
in some cases little or no acetylene could be found at the top of the 
inner zone of the flame, it either having diffused with the hydrogen 
and been consumed, or polymerised to other compounds. 

It is manifest that the luminosity of a flame will be governed, 
not by the percentage of acetylene in the gas, but at the point at 
which the temperature is sufficiently high to bring about decom- 

If, instead of making a mixture of 90 per cent, hydrogen and 10 
per cent, acetylene, the hydrogen is burnt at the end of an open 
platinum tube, which has a fine platinum tube passing up the centre 
to the top of the inner zone of the flame, and if the acetylene be 
passed into the flame at the rate of one volume for every ten of the 
hydrogen, not only do we obtain an intensely luminous, but a very 
smoky flame. 

In this experiment the gases were issuing from their respective 
tubes at the same pressure, but the small tube soon choked from 
deposited carbon, and it was found that the same results could be 
equally well attained by drawing down the inner tube to the level of 
the hydrogen tube, and making the acetylene issue at a slightly 
higher rate of flow, which hurried it in a compact stream through 
the inner zone of the hydrogen flame. 

In order to see if the percentage of acetylene present at the top of 
the non» luminous zone bore any ratio to the illuminating value of the 
mixture, experiments were made in which mixtures of hydrogen and 
acetylene were burnt at a small flat flame burner, and the percentage 
of acetylene was determined by gently aspirating out some of the 
flame gases from the top of the non-luminous zone. 

Analysis of mixture used. Illuminating 

^ — A ^ Acetylene at top of value of flame 

Hydrogen. Acetylene. non-luminous zone, for 5 cub. ft. 

66-5 34-5 372 14-0 

43-5 56-5 8-42 87-0 

O'O 100-0 14-95 240-0 


in the Flames of Ilydrocarhon Gases. 


On plotting out these resnltsj tliej certainly seem to point to 
tlie fact that, with flames of the same size burning from the same 
burner, the light emitted by the flame is directly proportional to the 
percentage of acetylene present at the top of the non-luminous zone 
of the flame, provided always that the temperature is sufiiciently 
high to complete its decomposition. 

li'iG. 2. 

' ■ — 

^ 16 

Is '^ 

^ ^ ^' 








- — 



- — *" 





- "^ 

= — . — -. 




^ '' 


Q^ ^ 10 20 50 40 50 60 70 60 90 100 200 240 
Catidlc powt.r calculated to a corisumjjti ofv of 5c f per hour 

It is perfectly possible for the temperature of a flame to be so little 
above the point necessary to decompose the diluted acetylene that, 
whilst some decomposes and renders the flame faintly luminous, the 
larger portion burns without decompositioQ. A good example of this 
is to be found in the combustion of alcohol, the flame of which con- 
tains as much acetylene as is to be found in a good coal-gas flame, 
but which is practically almost non-luminous. If alcohol in a small 
dish be ignited, it burns with a faintly luminous flame, and if a bell- 
jar is placed over it, some of the products of combustion mingling 
with the flame still further cool it and render it non -luminous ; but 
if now a stream of oxygen be introduced under the bell- jar the 
temperature of the flame is at once increased and becomes higbly 
luminous, whilst a cold porcelain vessel held in the flame is coated 

with soot. 

In all the experiments in which light was developed in heated 
tubes by the decomposition of acetylene, the glow of the carbon was 
red and lurid, the light emitted being of the same character and 
appearance as that developed by the combustion of potassium in 
carbon dioxide, and entirely lacking the pure white incandescence of 
the acetylene flame as burnt from a flat- flame burner. 

This may be due to the fact that in the open flame the tempera- 
ture of the carbon particles is presumably due to three sources of 
heat : — 

462 Prof. V.B.Lewes. The Cause of Luminosity [Mar. 21 5 

A. Heat derived from the decomposition of the acetylene molecule. 

B. Heat derived from the combustion of hydrogen, carbon mon- 

oxide, and some hydrocarbons in the flame. 

C. Heat derived from the combustion of the carbon particles 


whilst in the tube experiments the heat of the walls of the tube 
and the heat of decomposition alone are acting, and it is evident 
that the intensity of tbe beat finding its way through the walls of 
the tube will be very different to that exercised by the walls of burn- 
ing gas which enclose the luminous portion of the flame, and there 
can be but little doubt that the temperature of the carbon particles 
will vary enormously with the rate at which the acetylene decom- 
poses, as the more quickly the action takes place, the greater will be 
the localising action upon the heat evolved, and the higher the incan- 
descence of the carbon particles . 

That this is so seems certain from the whiteness of the flash of 
light emitted when the acetylene is detonated, and experiments 
were made in order, if possible, to gain an idea as to how much 
of the incandescence of the carbon particles was due to the 
endothermicity of the decomposing acetylene, and how much to 
the action of heat and combustion on the carbon particles after 

In order to do this, a non-luminous flat flame of large size was 
desired, and was obtained by using coal gas de-illuminated by slowly 
passing it through bromine, well washing; with sodic hydrate solution 
and water, and then passing it through strong sulphuric acid, the 
gas so treated having an illuminating value of 1'2 candles for 5 cubic 
feet when burnt in the London argand at such a rate as to give a 
3-in. flame, whilst in a fish-tail burner it gave a non-luminous 
flame. This gas gave on analysis the following percentage compo- 
sition : — 

Carbon dioxide 0*00 

Unsaturated hydrocarbons . . 0*00 

Carbon monoxide 5*50 

Saturated hydrocarbons. ..... 83*28 

Hydrogen 55*25 

Nitrogen , 5*49 

Oxygen , 0*48 


so that its combustion would give practically the same temperature 
and flame reactions as those in an ordinary gas flame. 

A very fine platinum tube was now obtained, closed at one end, and 
with fiYQ minute holes bored in a line close to the sealed end, and 

1895.] in the Flames of Hydrocarhon Gases. 463 

this jhaving been so arranged that the holes were buried in the flame 
just at the top of the inner zone, acetylene was then gently allowed 
to flow through them into the flame. 

At the points where the acetylene issued into the flame, small areas 
of intense luminosity were produced, whilst the liberated carbon 
streaming up between the flame walls of the upper zone produced 
dull red bands of very low luminosity. It may be suggested that 
the carbon particles supplied in this way to the flame may have 
agglomerated and formed masses larger than those produced in the 
ordinary way, but I do not think this, as they were completely 
consumed, and no smoke escaped from the crown of the flame, 
whereas if a flat flame is interfered with in such a way as to cause 
the carbon particles to roll themselves together, smoking of the flame 
is produced. 

I think the inference to be drawn from this experiment undoubtedly 
is that it is the heat of decomposition which gives the high incan- 
descence and light emitting value to the carbon particles, and that 
the temperature of the combustion of the other flame gases and finally 
of the carbon itself plays but a secondary part. 

In considering these results, it seems remarkable that if acetylene 
owes its power of rendering hydrocarbon flames luminous to its 
high endothermic properties, that cyanogen, which is still more 
endo thermic, should burn under all conditions that have at present 
been tried with a non-luminous flame. 

Heat of formation. 
Acetylene ........ C2H2 -—47,770 

Cyanogen C^^h —65,700 

It is clear that if the rapidity of decomposition localises the heat 
evolved to the products of decomposition, and that this renders the 
liberated carbon particles incandescent, whilst the hydrogen plays at 
best a very subsidiary part, it ought not to matter whether it be 
hydrogen or nitrogen which is combined with the carbon. 

Berthelot showed that cyanogen like acetylene could be detonated 
by a small charge of mercuric fulminate, but he notes that the test is 
not always successful, which points to the decomposition of this 
body requiring a greater expenditure of energy to break up the 
molecule than is the case with acetylene, and known facts would 
lead us to expect that this would be the case, as although exothermic 
compounds become less and less stable with rise of temperature, 
endothermic bodies on the other hand become more stable, 
and the endothermicity of cyanogen being greater than that of 
acetylene, would lead one to expect that temperatures which would 
decompose acetylene would have no effect on cyanogen, and that, 
as during the combustion of cyanogen, the liberation of nitrogen 

464 Prof. V. B. Lewes. The Cause of Luminosity [Mar. il, 

would probably have a dihiting and cooling action, the cyanogen 
would burn directly without liberating any carbon whicb could emit 

In order to see if the temperature of the cyanogen flame when 
burnt from an ordinary flat flame burner differed much from that 
of hydrocarbons when consumed in a flame of the same size and 
kind, the temperatures were experimentally determined by the same 
method employed, and in the same parts of the flame as had before 
been done with acetylene, ethylene, and coal gas. 

Portion of the flame. Temperature. 

Centre of inner zone 1877° 0. 

Top of inner zone ......,.»...«o...« 2085 

Near top of outer zone , ♦ 1645 

Showing that the cyanogen flame was actually hotter than the 
acetylene and ethylene flames, and about the same as the coal gas 
flame, but that the heat was differently distributed, the inner zone 
being far hotter than in the other gases, whilst the maximum 
temperature of the flame was at the apex of the inner zone, instead 
of being nearer the top of the flame. 

An experiment was now made to ascertain if it were possible to 
decompose cyanogen with luminous deposition of carbon, by passing 
it through a hard glass tube heated by means of the blowpipe ; but at 
the highBst temperature attainable no trace of any deposition of carbon 
took place, showing how far more stable cyanogen is under the influence 
of high temperatures than acetylene. 

The structure and characteristic appearance of the cyanogen flame 
have been explained by Smithells^ and Dent, who conclude that the 
inner zone of peach blossom tint is caused by the combustion of 
the cyanogen to carbon monoxide and nitrogen, whilst the outer blue 
cone is formed by the oxidation of the monoxide to dioxide, the green 
fringe to the outer cone being attributed to the presence of small 
quantities of oxides of nitrogen ; and if this explanation be accepted^ 
it is clear that we could not obtain luminosity in the portion of the 
flame immediately above the inner zone, as all cynogen has been 
destroyed without decomposition before that point is reached. It is 
conceivable, however, that although no luminosity can be detected 
in a cyanogen flame, and although the temperature which can be 
obtained in a glass tube is insufficient to break up the compound with 
luminous separation of carbon, yet if cyanogen could be heated to a 
considerably higher temperature, it might be possible to decompose 
it in such a way as to develop luminosity. 

In order to try this point, a hydrogen flame was burnt from the 
€nd of an open platinum tube 9 mm. in diameter, and a thin platinum 

* ' Chem. Soc. Jour.,' 1894, p. 603. 

1895.] in the Flame of Hydrocarbon Gases, 465 

tube 2*0 mm. in diameter was passed up tlirougli tlie broad tube to 
the apex of the inner zone, and a slow stream of cyanogen was 
admitted, with the result that the flame at once became luminous, 
and on surrounding the hj'-drogen flame with an atmosphere of 
oxygen to increase the temperature, the luminosity was considerably 

This experiment at once explains the cause of the non-luminosity 
of the cyanogen flame, and shows that it is purely a question of 
temperature, and thfe probabilities are that, burnt in a flame which 
gave sufficient heat to rapidly decompose it, nearly as high an 
illuminating value as that of acetylene would be obtained. 

I think the explanation of the apparent anomaly of the cyanogen 
flame having a higher temperature than the acetylene and ethylene 
flames, is to be found in the fact that the molecules of cyanogen 
are consumed without previous decomposition, so that the heat 
absorbed during the formation of the cyanogen is added to the 
heat of combustion, and raises the average temperature of the flame, 
whereas with acetylene the instantaneous decomposition of the 
molecule before combustion confines the heat evolved to the liberated 
products, and the average temperature of the flame is but little more 
than the heat of combustion. 

If the luminosity of a hydrocarbon flame is principally due to the 
localisation, during intensely rapid decomposition, of the heat of 
formation in the products, the illuminating values of such hydro- 
carbon gases as contain two atoms of carbon in the molecule should 
bear a simple ratio to their heat of formation. The gaseous hydro- 
carbons are — 

Heat formation at 
Hydrocarbon. Composition. constant pressure. 

Ethane OgHg +25670 

Ethylene O2H4 — 8000 

Acetylene O2H2 —47770 

and although, they may undergo many changes in the flame, they will 
all ultimately be reduced to carbon and hydrogen again before the full 
luminosity of the flame is developed. 

When the acetylene into which these hydrocarbons is converted 
by heat is decomposed, the action takes place with such enormous 
rapidity that one would expect the heat evolved to simply divide itself 
amongst the liberated atoms, so that the question of specific heat at 
high temperatures may be omitted. 

With exothermic compounds like ethane, considerable heat will 
have to be developed by its own combustion before it is converted 
into tbe acetylene, which, by its decomposition, endows the flame 
with luminosity, and if we take the ethane and call its light pro- 
ducing energy 1, we can then obtain a ratio of such energy for the 


466 Prof. V. B. Lewes. The Cause of Limiinosity [Mar. 21, 

other liydrocarbons available for distribution amongst tbe products 

of decomposition. 

Ethane 9^870 ^^ 

-r.. . 256704-8000 
Ethylene • « 9^i'^0 ~ Vol 

A . 1 25670 + 47770 ^, _ 
Acetylene 9^a'70 ~ "^'^ 

These ratios must now be divided amongst the atoms liberated at 
the moment of decomposition from the molecule, and we thus obtain 
the ratio :— 

1 . VU . 2;86 
8 ' 6 '4 

or 1 : 1-74 : 572 

The determination of the illuminating value of a gas becomes more 

and more difficult the higher its illuminating value, owing to the 

cooliug effect of the small burners that must of necessity be used in 

order to ensure complete combustion. Dr. Percy Prankland* assigned 

the illuminating value of 35 candles to ethane as the mean of four 

tests, which varied considerably amongst themselves, and, adopting 

his figure, the calculated illuminating values for the ethane, ethylene, 

and acetylene would be : — 

Illuminating value. 

Calculated. . Found. 

Ethane 1 x 35 =: 85 35 

Ethylene 1'79 X 35 = 60-9 68*5 

Acet^rlene 5-72 x 35 = 200*2 240 

figures which are far nearer the experimental ones than could have 
been expected, considering the crude character of the calculation and 
insufficient data, which leads to omitting altogether such important 
factors as the amount of gas consumed to bring about the requisite 
temperature of decomposition, the specific heat of the products, and 
the thermal value of the change from gaseous to solid carbon, and are 
of no value except as showing that a ratio does exist between heat of 
formation and illuminating value. 

Methane is the only other gaseous hydrocarbon of which the heat 
of formation is known, it being -4- 21750, and as the molecule contains 
only 1 atom of carbon, 2 mols. have to be taken, and on calculating 
the probable illuminating value by the same method as was applied 
to the other hydrocarbons, we should have — 

^ ' Chem. Soc. Jour.,' vol. 47, p. 237. 

1895.] in the Flame of Hydrocarbon Gases. 467 

26670 + {25670- (21760 X 2) }^oK 

26-670 '^-^ _ g.4 


and tlie illuminating value, as determined by Mr. Lewis T. Wriglit, 
is 5*2 ; but here, again, we know by experiment that methane requires 
a Terj high temperature to bring about its conversion into acetylene 
and decomposition into carbon and hydrogen, and that a large 
portion of the gas must be burnt without decomposition to do 

The facts which I have sought to establish in this paper 
are : — 

1. That the luminosity of hydrocarbon flames is principally due to 
the localisation of the heat of formation of acetylene in the carbon 
and hydrogen produced by its decomposition. 

2. That such localisation is produced by the I'apidity of its decom- 
position, which varies with the temperature of the flame and the 
degree of dilution of the acetylene. 

3. That the average temperature of the flame due to combustion 
would not be suflicient to produce the incandescence of the carbon 
particles wdthin the flame. 

In my paper on the action of heat upon ethylene, brought before 
the E/Oyal Society this spring, I showed that the decomposition of 
ethylene into acetylene and simpler hydrocarbons was mainly due to 
the action of radiant heat, and was but little retarded by dilution, 
whilst I have shown in this paper that the acetylene so produced 
requires a considerable increase in temperature to bring about its 
decomposition when diluted, and it is possible with these data to 
give a fairly complete description of the actions which endow hydro- 
carbon flames with the power of emitting light. 

When the hydrocarbon gas leaves the jet at which it is being burnt, 
those portions which come in contact with the air are consumed and 
form a w^all of flame which surrounds the issuing gas. The unburnt 
gas in its passage through the lower heated area of the flame under- 
goes a number of chemical changes, brought about by the action, of 
radiant heat emitted by the flame walls, the principal of which is the 
conversion of the hydrocarbons into acetylene, methane, and hydro- 
gen. The temperature of the flame quickly rises as the distance from 
the jet increases, and a portion of the flame is soon reached at which 
the heat is sufficiently intense to decompose the acetylene with a 
rapidity almost akin to detonation, and the heat of its formation, 
localised by the rapidity of its decomposition, raises the liberated 
carbon particles to incandescence, this giving the principal part of 
the luminosity to the flame ; whilst these particles, heated by the 
combustion of the flame gases, still continue to glow, until final ly^ 

2 L 2 

468 On the two-fold Spectra of Oxygen and Nitrogen. [Mar. 21^ 

themselves consumed, tliis external heating and final combustion 
adding slightly to the light emitted. 

Any unsaturated hydrocarbons which have escaped conversion into 
acetylene before luminosity commences, and also any methane which 
may be present on passing into the higher temperatures of the lumin- 
ous zone, become converted there into acetylene, and at once being 
decomposed to carbon and hydrogen, increase the area of the light- 
giving portion of the flame. 

My thanks are due to Mr. F. B. Grundy for the help he has given 
me in the work entailed by this paper. 

IV. " A possible Explanation of the two-fold Spectra of Oxygen 

and Nitrogen." By E. C. C. Baly, A.I.C, 1851 Exhibition 
Scholar in University College, London. Communicated: 

by Professor Eamsay, F.R.S. Eeceived February 27^ 1895. 


The two spectra of oxygen are shown to be of a different nature. 
They behave differently, and reasons are given for their being in 'all 
probability the spectra of different gases. They may either be two 
spectra produced by different vibrations of the oxygen molecule, or 
they may be the spectra of two different modifications of oxygen, or 
the spectra of two distinct gases resulting from a dissociation of 
oxygen, a combination of whicli is called oxygen. 

It appeared worth while to undertake experiments with a view of 
testing the last of these. Oxygen was sparked in an apparatus 
similar to that used by Professor J. J. Thomson in his experiments 
on the electrolysis of steam. Hollow platinum electrodes were used, 
each one of which was connected with a Sprengel mercury pump. In 
the first experiments, the distance betw^een the electrodes was 35 mm.y 
and the highest pressure compatible with the appearance of the twQ 
spectra was made the starting point of the experiments. In these 
first experiments it was 380 mm. The density of the oxygen before 
sparking was determined, and taken as a test of its purity. The 
fractions obtained from the anode and cathode were weighed, and the 
results are given. They follow the lines of J. J. Thomson's results, 
inasmuch as with long sparks a lighter fraction was obtained at the 
cathode, and with short sparks a heavier fraction. The fractions from 
the anode were not so definite as from the cathode, though the differ- 
ence was in the right direction. The probable maximum error of 
weighing was 0*0001 gram. This meant exactly one in the second 
decimal place of the density obtained. The general accuracy of 
the results may be gauged from the densities of unsparked oxygeB 


'■ "^'- -■